Before you pay anyone — me included — to rewrite your website, run these eighteen checks yourself. Most take under five minutes. Several will find money you are currently leaving on the table, and a few will save you from commissioning a rewrite you did not need.
Work through them in order on your homepage first, then repeat for your highest-intent page (usually a service or pricing page). Score as you go; there is a scoring guide at the end.
How to use this audit
Three ground rules make this worth doing properly.
Use a real device and real conditions. Open your site on your phone, on mobile data, in a private window. Not on your 27-inch monitor with everything cached. Around two thirds of visitors to most service sites arrive on a phone, and the phone experience is the one that counts.
Get someone else to do half of it. Checks 1, 2, 6 and 16 are useless when performed by the person who wrote the copy. You know too much. Ask someone outside your business — ideally someone who fits your customer profile.
Write down evidence, not verdicts. “Homepage is confusing” is not actionable. “Three of five testers could not say what we sell after five seconds” is.
Part one: clarity (checks 1–5)
Check 1 — The five-second test. Show your homepage above the fold to five people for five seconds, then hide it. Ask: what does this company do, and who is it for? Pass: three of five correct on both. This single check predicts more about your conversion rate than any other on the list.
Check 2 — The competitor swap. Paste your headline onto your closest competitor’s homepage. If it still reads as true, your headline is a category description, not a position. Pass: the swapped headline is visibly wrong.
Check 3 — The jargon sweep. Search your homepage for: solutions, innovative, passionate, seamless, cutting-edge, best-in-class, bespoke, holistic, journey, empower, leverage, robust. Pass: fewer than two hits. Each one is a word your customer did not use in any interview you have ever run.
Check 4 — The audience-name check. Can you find, in the first screen, an explicit statement of who this is for? Not implied — stated. “For allied health clinics with 2–15 practitioners.” Pass: yes, and it names a segment narrow enough to exclude someone.
Check 5 — The read-aloud test. Read your hero copy out loud to a colleague. Note every place you stumble, rephrase or add “well, basically what we mean is…”. Pass: no stumbles, no clarifications. The sentence you spontaneously add is almost always better than the one on the page.
Part two: structure (checks 6–9)
Check 6 — The headings-only skim. Hide all body copy — use your browser’s reader mode or simply scroll fast. Read only headings and buttons. Pass: the story still makes sense and a reader could act on headings alone. Around half of visitors read essentially nothing else.
Check 7 — One job per page. Write down, in one sentence, what a visitor should do on this page. Then count the calls to action. Pass: one primary CTA, visually dominant, with any others clearly secondary. Two equal buttons is a coin toss you outsourced to the customer.
Check 8 — The scroll-depth reality check. Open your analytics and find where people stop scrolling. Now look at what is immediately above that point. Pass: your key proof and primary CTA both sit above the point where 50% of visitors stop. The most common finding in my audits is excellent proof buried below the fall-off line.
Check 9 — Paragraph length on mobile. On your phone, find the longest unbroken block of text. Pass: no paragraph exceeds four lines on a phone screen. A paragraph that looks tidy on desktop can be a fourteen-line wall on a phone, and walls do not get read.
Part three: proof (checks 10–12)
Check 10 — The claim-to-proof ratio. Highlight every claim on the page. For each, is there supporting evidence within one screen of it? Pass: every claim of superiority, speed or results has adjacent proof. Unsupported claims do not merely fail to persuade — they actively discount the claims around them.
Check 11 — Testimonial specificity. For each testimonial, check three things: full name, role and company; a specific before-and-after; and relevance to your target segment. Pass: at least three testimonials with all three. “Great service, highly recommended — J.S.” is decoration.
Check 12 — The objection audit. List the top five reasons people do not buy from you (ask your sales team, or your own memory of lost deals). Now find where each is addressed on the site. Pass: at least four of five answered explicitly, at the point in the page where the doubt arises — not exclusively in a FAQ at the bottom.
Check 12 usually finds two or three unanswered objections. Adding a short, honest paragraph answering each — right where the doubt occurs — is typically an hour of work and the single highest-return change most sites can make.
Part four: search (checks 13–15)
Check 13 — Title tags. View the browser tab, or check the page source, for every important page. Pass: each is unique, under about 60 characters, front-loads the words a customer would search, and is not “Home | Company Name”.
Check 14 — The intent match. Search the main term you want the page to rank for. Look at what actually ranks. Pass: your page format matches the dominant format on that results page. A service page competing against a page of guides will not win, however good it is.
Check 15 — Internal links. Count the descriptive internal links on your key pages, and check where your best article points. Pass: at least three contextual links per page, with descriptive anchor text, and every high-traffic article links to a relevant service page from inside the body copy.
Part five: conversion (checks 16–18)
Check 16 — Button language. Read each button label in isolation. Does it complete the sentence “I want to…” from the customer’s perspective? Pass: every label is a verb phrase naming the outcome or next screen. “Submit”, “Learn more” and “Click here” all fail.
Check 17 — Friction at the ask. Count the fields on your main form and the number of unknowns at the point of clicking. Pass: no more than six fields for a first enquiry, a line of microcopy setting expectations for what happens next, and no surprise required fields. Every additional field costs completions; every one must earn its place.
Check 18 — The price question. Time how long it takes a first-time visitor to find any indication of cost. Pass: under ten seconds to reach at least a starting price or range. “Contact us for pricing” is a filter that removes serious buyers as efficiently as tyre-kickers, and it is the check businesses argue with me about most.
Scoring your site
| Checks passed | What it means | What to do |
|---|---|---|
| 15–18 | Your copy is working. Gains from here are incremental. | Focus on traffic and offer, not rewriting. |
| 11–14 | Solid foundations, specific leaks. | Fix the failed checks individually. Do not rewrite. |
| 7–10 | Structural problems. Patching will not hold. | Rework the message hierarchy, then the pages. |
| 0–6 | The site is arguing for someone else’s business. | Start with research, not writing. |
What to fix first
Order matters more than volume. Working in this sequence means each fix makes the next one easier:
- Check 1 and 2 — the hero. Everything downstream inherits the clarity or confusion of your first screen. Nothing else is worth doing until this passes.
- Check 12 — unanswered objections. Cheapest meaningful win available. An hour of honest writing.
- Check 7 and 16 — one CTA, said properly. Structural, quick, and immediately measurable.
- Check 11 — real testimonials. Requires asking three customers three questions. Do it this week.
- Check 18 — price transparency. Usually a business decision rather than a copy decision, which is why it takes longest. Start the conversation now.
- Checks 13–15 — search. Slower to pay off, so start early, but never before clarity is fixed. Ranking a confusing page just means more people bounce off it.
Run the audit twice a year, and always before you commission a redesign. A surprising number of “we need a new website” conversations turn out, on inspection, to be “we need four paragraphs rewritten and the price to be findable” — which is a far better afternoon’s work than a six-month rebuild.